OpenAI and EU in talks over cyber model access; Anthropic blocks Mythos deployment

OpenAI is negotiating with the EU over deployment terms for a specialized cybersecurity model, while Anthropic remains a holdout—refusing to commit to EU-preferred access controls on its Mythos model. The standoff reflects tension between Brussels' foundation model governance requirements and US vendors' data residency and IP protection concerns.

Regulatory gating is becoming a material factor in product release schedules. The EU's regulatory posture on frontier models continues to shape competitive dynamics between Anthropic (cautious) and OpenAI (negotiating alignment).